采用稀疏测量的高分辨率相干声场分离

现有相干声场分离方法受限于采样定理,要获取高分辨率的分离结果需要大量的测点。为降低测量成本,提出一种采用稀疏测量的高分辨率相干声场分离方法。该方法首先根据波叠加原理构建传递矩阵,然后通过奇异值分解技术获取声场的一组稀疏正交基,最后利用稀疏正则化获取权重系数的稀疏解并将其用于分离目标声场。仿真和实验结果表明:在测点数较少的情况下,该方法相较于常规等效源声场分离方法的分离精度更高。同时,常规等效源声场分离方法的分离结果分辨率受限于测点数目,而该方法可以在测点数较少情况下实现高分辨率的相干声场分离。     

南中国海存在孤立子内波条件下的声场时间相关半径

在动态的海洋环境中,由于数据向量和拷贝场之间的失配,匹配场处理器的性能会发生退化。数据向量的时间相关半径是这种退化的一种量度。通过2001年ASIAEX南海实验中垂直阵上水听器接收到的声场数据求取了400 Hz窄带信号的声场时间相关。从实验数据处理结果观察到,伴随着传播路径上非线性内波的进入,声场的时间相关半径减小。同时利用一个二维的平流冻结海洋模型和传播路径上三个温度链的温度数据对声场进行了数值仿真,分析了不同频率下的声场时间相关半径。结果表明:实验结果与仿真的400 Hz信号的声场时间相关较为一致。可见,在时变的海洋环境下,声信道中存在孤立子内波将会使声场的时间相关半径大大缩短。     

基于相关子空间的双峰发射线光谱特征提取与分析

低分辨率天体光谱中呈现的双峰发射线轮廓可能反映着比较珍稀的物理现象,如双活动星系核（AGNs）、双超大质量黑洞（SMBH）、喷流云与窄线区的相互作用,也可能是光谱处理过程产生的低质量特征线。双峰发射线轮廓一般由2个或2个以上的波峰构成,对该特征进行提取并分析,可作为双AGN、星系对、双黑洞等稀有天体更有效的搜寻依据,从而有助于更深入地研究星系乃至宇宙的形成与演化。提出一种新的基于相关子空间的双峰发射线特征提取与分析方法,主要工作分为以下三部分：首先利用稀疏差异因子δ度量双峰发射线光谱中属性差异程度,利用KNN方法约束参与稀疏差异因子计算的光谱范围,在此基础上,针对LAMOST低分辨率光谱给出基于相关子空间的特征提取方法;其次,为了验证稀疏差异因子σ以及KNN的输入参数k对双峰发射线光谱的适应性,选择LAMOST双峰发射线光谱样本及普通星系光谱数据,红移范围z<0.3（确保Hα,Hβ,[OⅢ]λλ4 959, 5 007,[NⅡ]λλ6 548, 6 584,[SⅡ]λλ6 717, 6 731等发射线落在LAMOST波长覆盖范围）,获得训练集光谱总数332+332（正负样本）条,并借助人眼检查分析该方法中的两个参数：k和δ阈值α对结果的影响,实验表明,当k=18,α=0.6时,相关属性分布较密集且稀疏点较少,结果比较理想;最后,对332条双峰发射线光谱特征子空间所在的波长区间、双峰红/蓝移间隔、双峰线强比等进行了理论分析、测量及人眼认证的基础上,给出了基于相关子空间的双峰发射线特征表述。此外,从双峰特征子空间上不同的发射线激发机制（[OⅢ]/[NⅡ]/[SⅡ]等禁线、Hα、Hβ等氢线）及相关特征子空间上线强关系等角度,对样本中双峰轮廓进行了分析。     

基于声线跟踪法对闭空间声场预测的接收球半径计算

接收球半径是采用声线跟踪法进行声场预测的一个重要参数。以前所用到的接收球半径一般是通过经验估计得到的,缺乏详细的理论分析。本文从声波传播机理角度分析了根据声线密度来确定接收球半径的原因,并推导了矩形闭空间中声线密度和接收球半径的计算方法。声线密度可以通过原始声线数目,声场空间体积与形状,边界吸声系数来确定。在一给定闭空间里,声线密度可看作是均匀分布的,所以接收球半径与空间位置无关,可看作是一个常数。但对不同的声场空间来说,由于空间体积、形状和吸声系数的变化,声线密度是不同的,因此接收球半径也会不同。声线密度越大,接收球半径越小;声线密度越小,接收球半径越大。实验表明,所提出的接收球模型能用来较准确地预测闭空间里的声压级和混响时间等声学参数。     

基于灰度相关的高分辨率红外图像重建

介绍了提高空间分辨率的受控和不受控的微扫描法。基于空间灰度相关分布特性,提出了一种微扫描误差修正算法,该算法利用灰度相关特性匹配低分辨率图像的位置关系。为准确评估成像质量,利用高精度METS-L型红外测试系统对MTF进行了测试。测试结果显示,该算法能较好抑制微扫描位移误差引起的图像模糊,有效提高了空间分辨率。     

两端开口空间的声场研究

研究了两端开口空间中声压分布同封闭空间和长空间中的不同,比较传统赛宾公式、改进赛宾公式和声线法对这种空间的估计准确度,同时对其中出现非指数衰变曲线进行讨论,分析其产生的原因、影响因素和规律。文中使用声线法进行数值计算,并且通过实验来和数值结果比较,验证结论。研究表明:两端开口空间中的稳态声压分布与封闭空间和长空间明显不同,混响时间的计算不能再使用传统赛宾公式,声线追踪法无法完全准确地估计稳态声压;其中出现非指数衰变曲线的原因经分析为空间内各不同活跃度的小区域之间的声能量耦合造成,而能否被观察到取决于能否满足耦合空间理论中对于双斜率衰变曲线出现的要求,直观上只要观察点和声源距离相差不远,并且不都在空间中央便可观察到明显的双斜率衰变曲线。     

海洋内波对水下声场的影响

海洋内波是海水内部的波动之一,它与海洋表面的海浪运动有着非常不同的物理特征。海洋内波会引起海水温度剖面发生剧烈的波动,这种波动具有很强的方向性和时变性,对水下声传播影响巨大,并展现出一些很奇特的声波导效应(包括二维波导效应和三维波导效应),给水下目标的探测与定位以及水声通信等带来严重影响。海洋内波对声场的影响研究涉及到海洋动力学和水声学两个学科领域,文章重点介绍海洋内波对声场影响研究领域的相关进展,并对内波环境下典型声波导现象进行了讨论,最后对该领域未来的发展进行了总结和展望。     

基于核磁共振的子空间量子过程重构

基于三核自旋量子系统实现了内嵌两量子位子空间量子过程重构.通过输入完备态集合对执行的量子过程进行了表征.由于量子过程内嵌于子空间,使得重构实验所需输入次数及时间显著减少,同时实验尽量避免使用时间较长的J偶合演化,有效控制了系统的退相干。     

基于子空间分析的人脸识别方法研究

人脸识别技术是模式识别和机器视觉领域的一个重要研究方向,在众多人脸识别的算法中,基于子空间分析的特征提取方法以其稳定可靠的识别效果成为了人脸识别中特征提取的主流方法之一。本文对目前应用较多的子空间分析方法进行了研究,具体介绍了线性子空间分析方法：主成分分析（PCA）、线性鉴别分析（LDA）、独立主成分分析（ICA）、快速主成分分析（FastICA）等及非线性子空间分析方法：基于核的PCA （KPCA）等的基本思想及其在人脸识别中的研究进展,包括一些新的研究成果。此外,还应用orl及Yale B人脸库对几个基础的子空间方法进行了验证实验。实验结果表明,在几个子空间分析方法中,FastICA算法取得了最高的识别率。最后结合实验结果对各算法的优缺点进行了分析总结。     

高分辨率空间相机的工程分析

以大型结构设计软件SolidEdge建立了高分辨率空间相机的精确的几何模型, 在此基础上用大型有限元分析软件MSC.Patran建立了相机的有限元模型, 并利用MSC/Nastran对该相机整机进行了工程分析. 其计算结果与试验结果相差10%以内.     

The effect of spatial coherence of sound field on subspace-based high-resolution spatial spectrum estimation

When the signals received by any two array elements are not fully correlated, the performance of DOA (Direction Of Arrival) estimation methods might degrade if the plane-wave model assumption is still used. For narrowband uniform linear array, by utilizing the property of Toeplitz matrix, it is proved that the received signals can be expressed as the sum of several plane-wave signals, which can be viewed as the "splitting" parts of the true sources. The "splitting" of sources makes it difficult to estimate the true number of sources. As a result, spurious peaks might occur in the spatial spectrum when subspace-based high-resolution spatial spectrum estimation technique is used.     

Spatial-temporal correlation of a diffuse sound field

The spatial correlation function of the sound in a diffuse field is a quantity widely used in many reverbrant room acoustic applications. Although results for the spatial and temporal correlation for pure-tone and band-limited diffuse fields have already been developed, these have not been generalized for other signal types. This work presents a generalized derivation of the diffuse field spatial-temporal correlation which can be used for stationary random signals with given power spectral density. It is shown that the spatial-temporal correlation depends entirely on the temporal correlation of the signal exiting the diffuse field, or alternatively on its power spectral density. A simulation using the plane wave model is presented for tonal and broadband diffuse sound fields.     

The near field correlation spectrum of a metallic film

The spatial two-point coherence function of the thermal radiation field at subwave-distance from a metallic film is calculated within fluctuation electrodynamics. Long-range oscillations as a function of lateral separation are seen; we interpret their features by comparing to the dispersion relations of surface plasmon polariton modes on the thin film. The hybridization of the surface modes on the top and bottom interfaces leads to a beatnote in the field correlations, for specific detector setups.     

The application of source methods in estimation of an interior sound field

In earlier studies, one has successfully developed three different source methods (SSM, similar source method; IPSM, internal parallel source method; ISM, internal source method) to estimate radiation and scattering sound fields. All these methods are estimations of exterior sound fields. In the present study, the similar source method is modified to estimate an interior sound field. The modification is to move the imaginary sources outside the boundary surface. In addition, general boundary conditions in terms of acoustic admittance or impedance are considered by introducing the suitable least square error functions. Finally, a two-dimensional interior sound field with alternative boundary conditions is evaluated to simulate sound fields inside a car. The results are in agreement with those of boundary integral method.     

Statistical estimation of the spatial spectrum of secondary sources

A statistical problem of estimating the shape and maximum possible width of the spatial energy spectrum of secondary sources is formulated and solved. The secondary sources under consideration are characteristic of the class of scatterers with preset statistical mean parameters. This estimation is important for analyzing the possibility of solving inverse scattering problems, including the question of the uniqueness and stability of the solution and also the correct organization of a tomographic experiment. Results of a computer simulation are presented, and the expected statistical mean spectrum is compared with a particular realization of the spectrum. In the case of a strongly pronounced rescattering, the statistical mean width of the spectrum is mainly determined by the effective velocity contrast in the ensemble of scatterers and depends only weakly on the scale of spatial fluctuations of the scatterers’ characteristics.     

Down-mixing of multi-channel audio for sound field reproduction based on spatial covariance

This article describes a method for automatic down-mixing multi-channel audio content on the basis of spatial covariance. Such a down-mixing method should be able to convert the signal of the original multi-channel audio system into that for an alternative system with the lesser number of channels, while maintaining the spatial impression of sound. Moreover, it should take into account the listener’s position and transfer function. Wave surface control and convolving the head related transfer function are techniques used in sound field control or reproduction. We consider that the spatial impressions of a sound field, which we perceive through our ears, are reproduced by preserving the relative relationship between observation points even if the wave surface is not completely controlled. Takahashi et al. proposed a new sound field reproduction method that we named “SOund field Reproduction based on sPAtial Covariance” (SORPAC). SORPAC can control the point-to-point covariance in a sound field. We expect that this sound field reproduction method based on spatial covariance can be applied to down-mixing of multi-channel content because SORPAC does not require the listener’s position or transfer function. This article describes SORPAC and its characteristics. We used SORPAC for down-mixing audio content. We confirmed that SORPAC-based down-mixing could accurately reproduce the interaural cross correlation (IACC) in relation to the listener’s position.     

声场分布中的负向声强

本文讨论了相距很近的两个点声源的声强分布,发现位于声源之上的平面法向声强在某些区域为负向声强。对纸盆扬声器声场的实测结果也显示出与理论计算相符的负向声强。本文的理论计算及实际测量结果对负向声强的传统解释增加了新的内容。     

The high-resolution microwave spectrum of cyclopropyl cyanide

The quadrupole hyperfine structure of six selected rotational transitions of cyclopropyl cyanide has been analyzed. The derived coupling constants are χ_aa = ?3.453(11) MHz and χ_bb = 1.759(16) MHz. The analysis yielded improved center frequencies for the multiplets and, in turn, led to slightly revised rotational constants: A = 15786.068(22), B = 3465.112(7), C = 3286.252(10) MHz; D_N = 1.18(30) kHz.     

Decimative subspace-based parameter estimation methods of magnetic resonance spectroscopy based on prior knowledge

The method Hankel Total Least Squares (HTLS)-PK, which successfully incorporates prior knowledge of known signal poles into the method HTLS, has been proven to greatly improve the performance for parameter estimation of overlapping peaks of magnetic resonance spectroscopy (MRS) signal. In addition, decimation is also proposed as a way to increase the performance of subspace-based parameter estimation methods in the case of oversampling. Taking advantage of decimation in combination with prior knowledge to estimate the MRS signal parameters, two novel subspace-based parameter estimation methods, HTLSDSumPK and HTLSDStackPK, are presented in this paper. The experimental results and relevant analysis show that the methods HTLSDSumPK, HTLSDStackPK and HTLS-PK are slightly better than the method HTLS at low noise levels; however, the three prior-knowledge-incorporating methods, especially the method HTLSDSumPK, have much better performance than the method HTLS at high noise levels in the terms of robustness, estimated accuracy and computational complexity. Even if some inaccuracy of prior knowledge is considered, the method HTLSDSumPK also shows some advantages.